Hallie Trial
Rice University
Houston (29.7° N, 95.3° W)
In the late 1800s, tissue-staining dyes began to captivate physician-scientist Paul Ehrlich. In particular, he noticed that some dyes stained only bacterial cells while leaving mammalian cells their natural color. Erhlich envisioned a toxic dye molecule that would bind to and kill bacteria but leave human cells perfectly healthy; such a molecule would cure bacterial illness. After testing more than 600 arsenic-based dye compounds, Ehrlich discovered the first modern antibiotic, arsphenamine. This compound allowed doctors to treat syphilis effectively for the first time.
Just like Ehrlich’s dyes, all drugs operate by selective binding: they stick to some molecules but not others. Penicillin, for instance, attaches to and inactivates a vital protein that bacteria need to synthesize their cell walls, but the compound is harmless to human proteins. Aspirin helps control pain and swelling by binding to the enzyme that produces hormones called prostaglandins and thromboxanes, which serve roles in inflammation.
With the ever-advancing modern understanding of biochemistry, scientists can pinpoint exact proteins involved in causing disease. Our current approach to drug development is to select a druggable target protein and then maximize the binding of a drug to that single protein while preventing other interactions. Ideally, minimizing effects on other proteins minimizes side effects. In the complex, intricately connected systems of living organisms, however, there are often multiple redundant pathways to achieve the same purpose. This means that when a drug alters the activity of just one biological molecule, other molecules can sometimes fill in, and the medication may not produce as strong an effect as expected. Furthermore, biological molecules influence one another’s activity in complicated ways, so unexpected side effects can emerge.
A new approach to drug design called network pharmacology considers the interconnected living system as a whole rather than focusing on just one druggable protein. Dr. Anil Korkut from MD Anderson Bioinformatics and Computational Biology explained that network pharmacology “focuses on how pharmacological agents can alter the molecular network within a cell.”
Molecules in a cell interact much like people within a large corporation. In order to determine which figures are the most important for the functioning of the corporation as a whole, one might consider who talks to whom, how their communication influences each person, and whether people could still get the information and orders they need without each communication exchange. Similarly, to investigate which molecules to target with drugs, molecular pharmacologists must use computational models to map the interactions between many different biological molecules and the pathways relevant to a disease. They might find one single, vital “CEO” molecule that they can target with a drug to alter the whole disease “corporation.” Other times, they might find a collection of a few important “employee” molecules that can alter the direction of the disease “corporation” when targeted together but not when working alone.
Scientists have investigated network pharmacology through experimental procedures like systematic screening and computational network analysis. In systematic screening, researchers study the relationship between only two drug targets at a time. For example, they might have one anticancer drug that they wish to combine with another drug—a co-drug—to improve its effectiveness. They could add the first anticancer agent to thousands of tiny cancer cell cultures, add a different second drug to each culture, and determine which two-drug combinations work most effectively. Alternatively, scientists can perform synthetic lethality chemical sensitization screening. This requires them to create many different cancer cell lines, each with one gene turned off. Then, they add their drug of interest to each cell line. Some cell lines will show increased sensitivity to the drug, and this tells investigators that the particular genes turned off in these sensitized cells might be useful targets for co-drugs. Screenings can often help uncover complicated, sometimes counterintuitive drug interactions and lead to improved medication combinations.
Much of the future of network pharmacology lies in applying computational methods to biological networks and the chemicals that alter them. Advancements in computer models of the molecular networks within cells improve the efficiency of screenings by helping identify in advance which target combinations will likely effect the desired change. Once researchers decide which proteins or molecules to drug, cheminformatics databases correlating molecular structure with biological function and molecular binding can help chemists design compounds that affect those targets.
An inherently interdisciplinary field like network pharmacology demands researchers from many different academic backgrounds. Projects need computationally-trained individuals with some knowledge of chemistry, biochemistry, or both, as well as experimental chemists and biologists with enough understanding of computational models to apply them. Most positions require a PhD in a related discipline, such as computational biology or chemistry, systems biology, statistics, biochemistry, chemistry, and many others. Undergraduate students can prepare for these diverse graduate programs with a wide array of majors, including bioengineering, biochemistry, chemistry, computational and applied mathematics, statistics, and even biological physics.
A few organizations currently specialize in network pharmacology. Aside from MD Anderson where Dr. Korkut runs his lab, he has highlighted groups at Harvard Medical School and Oregon Health & Science University. Harvard has a specialized Laboratory of Systems Pharmacology, and OHSU houses the Center for Spatial Systems Biomedicine, whose researchers conduct many projects in the related field of perturbation biology. As network pharmacology develops as a discipline, more opportunities may emerge both in industry and in academia.
With further advancements in computational tools, network pharmacology may fundamentally alter how we think about chemical treatments for disease. It may also someday play a part in personalized medical care. If we develop the ability to create a distinct network model for each patient’s unique biology, Dr. Korkut stated, “we will be able to predict the immediate molecular changes in a given patient after therapy, and that will enable us to come up with better therapies, especially drug combinations… we are going to change people’s lives.”
Related Articles
This discusses the definition of network pharmacology, the goals of the field, relevant investigation platforms in the discipline, and future challenges. https://www.ncbi.nlm.nih.gov/pubmed/18936753.
This provides an introduction to networks in biology. https://www.ebi.ac.uk/training/online/course/network-analysis-protein-interaction-data-introduction/network-analysis-biology-0
The ACS page on cheminformatics provides information about the discipline cheminformatics, the questions it addresses, and careers available in the field. https://www.acs.org/content/acs/en/careers/college-to-career/chemistry-careers/cheminformatics.html
This research article authored by Anil Korkut and others provides an interesting example of network biology in action. The researchers constructed network models based on numerous experimental measurements to identify potential drug combinations for treating a drug-resistant melanoma cell line. https://elifesciences.org/articles/04640
Excerpts from a Longitude.site interview with Dr. Anil Korkut in student reflections.